A ‘natural’ choice for Nobel

The Nobel Prize in Economics this year has been awarded to Joshua Angrist and Guido Imbens. The Economic Sciences Prizes Committee believes that Card’s work has ‘challenged conventional wisdom leading to new studies and additional insights’.

The three economists together have ‘revolutionized empirical work’ in economics through their unique skills at spotting unintended ‘natural’ experiments. Eh? Natural experiments? Now what new stuff are the economists cooking?

There is a fair bit of history here, for which we need to travel in time to 1983, wherein Edward Leamer wrote that sublime paper titled ‘Let’s take the Con out of Econometrics’ published in the American Economic Review .

Leamer talks of a farmer who notices that there are areas in his field where birds roost in the trees and the yield under those trees is higher. He concludes that bird droppings increase the yield. But when he presents his findings in a scientific conference, he meets another farmer who says that he had thought that it was the shade that increased the yield. A young chap in the conference says wisely that the two hypotheses are really quite the same and that we are facing an ‘identification problem’. The nerds at the conference don’t quite understand this, but it sounds quite grand and becomes the go-word for the next few debates.

Contrast this to another situation wherein a planned exercise is undertaken to stake out plots within the land. Fertiliser is applied to some randomly selected plots and it isn’t applied to some other plots. Yield are calculated. The difference in the mean yield of the treated plots and the untreated ones is positioned as the effect of the fertiliser on the yields. This is a randomised experiment, with power to create an unbiased estimate of the effect of the fertiliser, thereby taking the ‘con’ out of econometrics. Such experimental evaluation to understand causalities became the hallmark of the work of quite a few economists at Princeton and Harvard in the eighties. The different statistical techniques in this space – difference in difference, propensity scores, regression discontinuity models and creatively chosen instrumental variables – captured the imagination and effort of many a young economist.

Labour issues

Card was a labour economics enthusiast. One of the key questions in labour economics was whether immigration led to lower wages for the natives of a country/state. The ‘con’ way would be to choose a city with no immigration and to choose a city with immigration. The difference in the wages in both the cities could be positioned to be the effect of immigration. Why is this ‘con’? Well, because the choice of cities is not random! There might be a reason that the immigrant workers make a beeline towards a particular city!

Now, in the 1980s, 1,25,000 Cuban immigrants entered the US and settled in Miami, adding about 7 per cent to the labour force in Miami in less than six months.

In this case, the choice of the destination city was driven by geographical and feasibility considerations, rather than a conscious decision to move to a vibrant labour market. Card recognised this to be a ‘natural’ experiment. He went about now choosing a city that could serve as control and used the difference-in-difference method to arrive at his first round of rather controversial conclusions. His conclusion was that there was no major difference in the wage levels in Miami and in other cities. While the paper was highly debated and its results were not immediately accepted, there is no denying that the paper created a standard for undertaking evaluations based on natural experiments.

In 1994, Card and Krueger (of Princeton) saw another opportunity for a natural experiment. The minimum wages in New Jersey were increased whereas those in the neighbouring state of Pennsylvania remained at the earlier levels. This was as natural a treatment and control as they could have hoped for. Card and Krueger compared the employment in both states and found that raising the minimum wages had not really impacted the employment in New Jersey significantly.

The search for meaningful causalities is indeed the holy grail in Economics. So dark the ‘con’ of econometrics!

The author is a brave economist trying to laugh against the odds

COMMENT NOW